back to article Red Hat bolts the stable with RHEL 6.7

Enterprises are rarely in a rush to upgrade their operating systems – they want others to do the battle testing for them first. As is the way with Red Hat Enterprise Linux 7, released in June 2014. Red Hat has many customers on the “stable, proven and predictable” RHEL 6.xx and is living up to its 10-year support pledge with …

  1. Anonymous Coward
    Anonymous Coward

    Red Hat has many customers on the “stable, proven and predictable” RHEL 6.xx

    And many customers are keen to avoid systemd, gnome3 and many of the horrors that RHEL7 throws at you. Upgrading from 5 to 6 was a nice smooth step forward while leaving people in an environment that they felt familiar and happy with. Jumping to 7 can leave people disorientated and thoroughly confused with the impression that if you run server systems in a global environment that your needs somehow weren't high on the designers priorities of some components.

    RHEL7 boots nice a quickly, even if the POST and FW stages of many modern systems more than compensate for that, but when you almost never reboot, who cares.

    1. Steve Davies 3 Silver badge

      Re: Red Hat has many customers on the “stable, proven and predictable” RHEL 6.xx

      Server boot times

      For an non Windows server reboots are pretty rare these days. It always irks me that reviewers seem to take great store by the boot time for a server. IMHO this is one of the most useless measurements I can think of.

      I guess they are just lazy or too bone idle to do a proper job of reviewing the products.

      Anyway with increasing use of SSD's who gives a dam about boot times anyway.

      I would be more concerned about how my workload performs than how fast it boots.

      1. sysconfig

        Re: Red Hat has many customers on the “stable, proven and predictable” RHEL 6.xx

        "For an non Windows server reboots are pretty rare these days. It always irks me that reviewers seem to take great store by the boot time for a server. IMHO this is one of the most useless measurements I can think of."

        Indeed. If a reboot (or downtime) of a single server causes a problem, the people in charge ain't doing it right [1]. The narrative that a single server's downtime would result in downtime for a site or application belongs into the last century.

        [1] or their bosses try to save money in the wrong place, in which case we want reboot times to be significantly longer to highlight the problem!

        1. Sandtitz Silver badge

          Re: Red Hat has many customers on the “stable, proven and predictable” RHEL 6.xx

          "If a reboot (or downtime) of a single server causes a problem, the people in charge ain't doing it right"

          That doesn't apply to small businesses. To have uninterrupted service while rebooting requires some sort of clustering, shared storage and thus server administration is beyond bosses nephews skills and must be outsourced to someone with skills and expensive hours.

          I'm somewhat interested to know how long the server is booting up, including all the POST stuff. When you're rebooting a server remotely (without iLO or other means to monitor the boot process) some servers with lots of memory and storage and start-up scripts take like 10+ minutes to come back on-line. It's nerve-racking unless you actually know the time it takes for full service.

    2. Anonymous Coward
      Anonymous Coward

      Re: Red Hat has many customers on the “stable, proven and predictable” RHEL 6.xx

      "And many customers are keen to avoid"

      We're a KDE house and have been for a loong time, back to when it was the RH standard.

      There is a _lot_ of resistance to moving gnomewards, let alone the abomination that is gnome3 (I stick with gnome2 at home)

  2. Daniel B.
    Boffin

    systemd!

    The main reason we're all sticking to RHEL 6.x is systemd. You can say it, El Reg. It's no secret.

    1. Destroy All Monsters Silver badge
      Trollface

      Re: systemd!

      So it needs a thorough redesign?

  3. Richard Lloyd

    Just started testing CentOS 7

    I've been playing with a CentOS 7 VM at home and for the desktop, adding in the MATE environment via EPEL actually does provide a look and feel not too far away from GNOME 2. It's certainly the best migration route for CentOS 6 GNOME desktop users.

    From a sysadmin point of view, there are a fair number of changes - getting used to systemctl instead of init.d scripts/chkconfig takes a bit of time though (luckily, the service command works the same way on 6 and 7, although it's usually just a systemctl alias in 7).

    CentOS 7 certainly boots more quickly than 6 thanks to systemd, but I think my gripe is mainly with GRUB 2 to be honest. It introduces a level of complexity to the config that isn't really needed. Gone are the days of just editing grub.conf by hand sadly, which was simple and very obvious.

    1. Trevor_Pott Gold badge

      Re: Just started testing CentOS 7

      Gone are the days of just editing grub.conf by hand sadly, which was simple and very obvious

      Not just grub; RedHat seem to be keen on doing away with this all over. Which is batshit fucking bananas crazy.

      1. AdamWill

        Re: Just started testing CentOS 7

        We didn't write grub2. RHEL 7 uses it because no-one is interested in maintaining grub-legacy, and there's no viable alternative. No-one particularly loves grub2, but no-one seems to hate it enough to make something better.

        1. Trevor_Pott Gold badge

          Re: Just started testing CentOS 7

          A multi-billion dollar company could easily maintain grub-legacy. For that matter, it could preserve the concept that "every config must be editable as a flat text file" thusly producing a sane and rational distro. None of this neo-registry bull that's all the rage.

          Red Hat is perfectly happy to throw its weight around in order to try to own new markets, or become relevant in markets where it dropped the ball. It's not remotely so interested at throwing its weight around to help ensure important packages in Linux adhere to concepts like flat text files or "doing one thing and doing it well".

          Red Hat has let the inmates run the asylum and the result is the first Red Hat distribution in my entire career that I flat out refuse to work with.

          Still, Red Hat gave 20 years of solid awesome. It was probably insane of me to think that this would continue indefinitely (or at least for the duration of the rest of my career). I'm sad about the steaming pile that Red Hat has become, but I don't have the will to fight.

          Red Hat's resources are billions of times my own. The mad hatters that want to ruin Linux in order to build their own little empires of ego and hubris are more charismatic, wealthier and better connected than I. If I leveraged every single connection I have, called in every favour I am owed, used every last penny I could beg, borrow or steal my discontent would register upon Red Hat not at all.

          So, to put things fairly bluntly: fuck 'em.

          There are alternatives. I am investigating them. I can do literally nothing to even get Red Hat's attention, but if I throw the full force of my capabilities towards helping some of the alternatives succeed maybe I can help a truly open, community-focused and user-oriented distribution grow.

          When life hands you OpenOffice by Oracle, you make LIbreOffice. I hope enough other people agree that this needs to occur that, combined, we can make it happen.

          1. Anonymous Coward
            Anonymous Coward

            Re: Just started testing CentOS 7

            FWIW I'm off doing a virtual (?) walk-about. I can dimly see something through the fog.

    2. AdamWill

      Re: Just started testing CentOS 7

      You can still edit grub.cfg by hand in RHEL / Fedora; the scary warning telling you not to comes from upstream grub2, which expects the config file to be updated by grub(2)-mkconfig, which does indeed nerf manual changes. In RHEL/Fedora's system, though, the config file is updated (on new kernel installs and so on) by grubby, which does not nerf existing manual changes to the file.

      The grub2 file does look a bit scary at first but a lot of it you can pretty much ignore, if you're just wanting to change kernel boot parameters and the like you can just find the appropriate lines and append the parameters you want, just like before.

  4. Andrew Meredith

    Not so bad

    Those of you who have not been using any *UX OSs other than RHEL 6 and similar, will probably see RHEL 7 as a major departure and a pain in the behind. However, if you have A) been using and updating Fedora as well (that's had systemd and gnome-shell for ages) and B) Have also been using crusty ancient commercial *UX platforms (particularly AIX) are likely to see RHEL 7 as a minor step away from RHEL 6 and the differences to be trivial.

    I do understand the theological issues with systemd and others that have departed from the usual *UX standards. It is a crying shame that systemd, for example, has now consumed a bunch of other perfectly functional packages and seems to be just getting bigger and bigger.

    However, once you have got the hang of how systemctl works, it does just work. I do have to say that dumping the use of startup shell scripts and moving to a config file that contains the settings necessary and can parallelise the startup, is a good thing (tm).

    I guess .. Bear With It .. is the best advice I can give.

    1. Anonymous Coward
      Anonymous Coward

      Re: Not so bad

      Centos 7 does have my attention for the moment. The theological issues are important though as I've been around since Unix was born: small is beautiful a mantra in all my work, even in DOS and Windows. Shells? If I don't like/tolerate it, it's toast. BTW, csh and Emacs are my faves.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like